Running Antimicrobial protection of extrafloral nectar

Nectars are rich in primary metabolites and attract mutualistic animals, which serve as pollinators or as an indirect defence against herbivores. Their chemical composition makes nectars prone to microbial infestation. As protective strategy, floral nectar of ornamental tobacco contains ‘nectarins’, proteins producing reactive oxygen species such as H 2 O 2 . By contrast, pathogenesis-related (PR) proteins were detected in Acacia extrafloral nectar (EFN, which is secreted in the context of defensive ant-plant mutualisms). We investigated whether these PR-proteins protect EFN from phytopathogens. Five sympatric species ( Acacia cornigera, A. hindsii, A. collinsii, A. 60 farnesiana and Prosopis juliflora) were compared, which differ in their ant-plant mutualism. EFN of myrmecophytes, which are obligate ant-plants that secrete EFN constitutively to nourish specialised ant inhabitants, significantly inhibited the growth of four out of six tested phytopathogenic microorganisms. By contrast, EFN of non-myrmecophytes, which is secreted only transiently in response to herbivory, did not exhibit a detectable inhibitory activity. Combining two-dimensional SDS-PAGE with nanoLC-MS/MS analysis confirmed that PR-proteins represented over 90 % of all proteins in myrmecophyte EFN. The inhibition of microbial growth was exerted by the protein fraction, but not the small metabolites of this EFN and disappeared when nectar was heated. In-gel assays demonstrated the activity of acidic and basic chitinases in all 70 EFNs, whereas glucanases were detected only in EFN of myrmecophytes. Our results demonstrate that PR-proteins causally underlie the protection of Acacia EFN from microorganisms and that acidic and basic glucanases likely represent the most important prerequisite in this defensive function.


INTRODUCTION
Plants secrete nectar to attract mutualistic animals, which mainly function as pollinators in the case of floral nectar or as defenders against herbivores in the case of extrafloral nectar (EFN) (Simpson and Neff, 1981;Heil, 2008;González-Teuber and Heil, 2009a).
Because nectars usually represent aqueous solutions of mono-and disaccharides 80 together with amino acids, they are prone to infestation by microbial organisms. When being present in the nectar, fungi (González-Teuber et al., 2009), and yeasts (Herrera et al., 2009) in particular, can alter its chemical composition and thereby reduce its suitability for the plant's animal mutualists (Herrera et al., 2008). Moreover, several phytopathogenic organisms may use the nectar-secreting tissues as entries to infect other plant organs (Bubán et al., 2003;Farkas et al., 2007). Therefore, being an excellent growing medium for yeasts, fungi and bacteria, nectar requires an efficient antimicrobial protection.
Unfortunately, our knowledge of the means by which plants protect nectar from microorganisms is extremely limited. Although the first reports on nectar proteins date 90 back to the sixties and seventies of last century (Baker and Baker, 1975;Lüttge, 1961), most studies that considered the defensive function of nectar focused on secondary compounds such as alkaloids and phenols. These metabolites commonly protect nectar from consumption by nectar robbers (animals that feed on nectar without providing a mutualistic service to the plant, see Stephenson, 1981;Johnson et al., 2006) or limit the duration of pollinator visits (Kessler et al., 2008). Only during the last decade, a series of studies discovered defensive proteins in the floral nectar of ornamental tobacco (Nicotiana langsdorffii × N. sanderae) (Carter et al., 1999). In this species, floral nectar contains a limited array of proteins termed "nectarins". Nectarins serve the protection from microbial infestation through a biochemical pathway called the Nectar Redox 100 Cycle (Carter and Thornburg, 2004a), in which mainly three of the five nectarins are involved: NEC1, NEC3 and NEC5. NEC1 was characterized as a manganese superoxide dismutase (Carter and Thornburg, 2000), NEC3 has carbonic anhydrase and monodehydroascorbate reductase activity (Carter and Thornburg, 2004b) and NEC5 is a glucose oxidase that functions together with NEC1 in the production of high peroxide levels (Carter and Thornburg, 2004c): nectar of ornamental tobacco can accumulate up to 4 mM of hydrogen peroxide: concentrations that are clearly high enough to exhibit toxicity on microorganisms. Thus, the floral nectar of ornamental tobacco is kept free of microbes mainly via the production of small reactive oxygen species.
By contrast, a proteomic study on EFN of the ant-plant, Acacia cornigera, 110 revealed the presence of several pathogenesis-related (PR-) proteins (González-Teuber et al., 2009). Myrmecophytes (ant-plants) are constitutively inhabited by specialised ant species, which serve as a very efficient indirect defence against herbivores (Heil, 2008).
In the most specialised cases, both the ant and the plant depend on this interaction, which thus represents an obligate mutualism. In the EFN of A. cornigera, activities of chitinase, ß-1,3-glucanase and peroxidase were detected together with proteins similar to PR-1, osmotin-like proteins and thaumatin-like proteins (González-Teuber et al., 2009). Most of these proteins were, however, only investigated by tandem mass spectrometry and characterized via MS-BLAST searches. Because no activity assays had been performed, the presence of these proteins could as yet not be causally linked to 120 the protection of EFN from microorganisms.
The present study was conducted to determine whether the antimicrobial protection of Acacia EFN can be directly and exclusively allotted to the enzymatic activity of its protein fraction, which would contrast the protective strategy of this nectar from the one that has been described by Carter, Thornburg and colleagues. We also aimed at investigating whether Acacia EFN inhibits the growth of phytopathogens and thus can serve in the protection from infection by pathogens that may use nectaries to enter the plant (Bubán et al., 2003). We used four sympatric Acacia species and a closely related Prosopis species, which exhibit different types of ant-plant mutualism and therefore differ in their EFN secretion schemes (Heil et al., 2004) andcomposition 130 (Heil et al., 2005;González-Teuber and Heil, 2009b). The obligate myrmecophytes among Central American Acacia species secrete EFN constitutively at high rates and the EFN of these species possesses a much higher level of proteins and of antimicrobial defence than the EFN of congeneric non-myrmecophytes (González-Teuber et al., 2009). The non-myrmecophytes, by contrast, secrete EFN at lower rates and only transiently in response to leaf damage; this EFN contains few proteins but high levels of sucrose (Heil et al., 2005;González-Teuber et al., 2009).
We studied the EFN of the obligate myrmecophytes, Acacia cornigera (L.) Willdenow, Acacia hindsii Benth. and Acacia collinsii Safford and of the two nonmyrmecophytes, Acacia farnesiana (L.) Willdenow and Prosopis juliflora Swartz. 140 Bioassays were employed to detect inhibitory activities of the nectars against phytopathogens and in-gel assays were used to determine the presence and functionality of basic and acidic chitinases and glucanases. Size exclusion filtration and heating of the EFN was used to investigate whether the antimicrobial activity of EFN is exclusively caused by the protein fraction. The results demonstrate that the antimicrobial protection of Acacia EFN is caused by the fraction of enzymatically active PR-proteins and independent of small, soluble molecules: an observation that represents a new strategy by which plants can protect nectar from infestation by potentially deleterious microorganisms.

Inhibitory effects of EFN on phytopathogens
Bioassays were performed following the disk diffusion method (Fig. 1) to investigate the biological activity against phytopathogens of extrafloral nectar (EFN), which was collected from wild plants growing in the coastal area of the state of Oaxaca, México.
As target organisms we used six phytopathogenic microorganisms (the oomycete Phytophthora parasitica and the fungi Fusarium oxysporum, Verticillium dahliae, Alternaria alternata, Botrytis cinerea and Plectosphaerella cucumerina, see Table 1 for details). EFN of myrmecophyte species inhibited the growth of at least 4 microorganisms ( Table 2). By contrast, no inhibitory effects were observed for non-160 myrmecophyte EFN as well as for the sugar solutions, which were used as controls (Table 2, Fig. 1). Size exclusion filtration and separately testing the protein fraction (> 5 kD) and the fraction of small metabolites revealed that only the protein fraction of EFN of the myrmecophytes inhibited the growth of Phytophthora parasitica (Table 3). As a second, independent attempt, EFN was collected from the three myrmecophytes and half of every sample was boiled before being used in bioassays. In response to heating to 100°C for 15 min, the EFNs of all three species lost their activity against the fungi, Alternaria alternata, Fusarium oxysporum and Botrytis capsici (Table 4).

Protein patterns and identification of PR-proteins 170
One-dimensional SDS-PAGE analysis (Fig. 2) of myrmecophyte EFN revealed ca. 9 -15 clearly distinguishable major bands, which ranged from 10 to 50 kDa. By contrast, EFN of the non-myrmecophytes exhibited very few bands, which ranged from 25 to 50 kDa (Fig. 2). A two-dimensional gel electrophoresis followed by MALDI-TOF/MS (matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry) and nanoLC-MS/MS (nanoflow liquid chromatography-tandem mass spectrometry) was employed to characterise the proteomes of the EFNs of A. hindsii and A. collinsii and of the two non-myrmecophytes. The 2D gel analysis revealed over 60 protein spots in the EFNs of the myrmecophytes, whereas between five and seven spots were observed in the EFN of non-myrmecophytes (Fig. 3). Consistent with the patterns seen in the 1D-180 gels (Fig. 2), the molecular weights of the main proteins ranged in from 10 to 50 kDa for both A. hindsii and A. collinsii and were over 25 kDa in case of the two nonmyrmecophytes. NanoLC-MS/MS allowed the annotation of 47 protein spots from the EFNs of the myrmecophytes (Table 5), whereas due to the low amount of protein present in each spot, no protein identification was possible in the EFN of A. farnesiana and Prosopis juliflora. De novo peptide sequences were obtained from 5 spots in EFN of Prosopis, which are presented in Table S1. Peptides identifying proteins for both myrmecophyte plants, A. hindsii and A. collinsii, are compiled in Table S2.
In order to quantify the extent to which chitinases and glucanases contributed to the total amount of EFN proteins, the PD Quest 7.3.0 program (Bio-Rad) was used to 200 conduct a relative quantification by determining the volume of each spot as optical density (OD) multiplied by its area [mm 2 ]. For A. hindsii, glucanase proteins contributed ca. 52 % + 2.1 to the total proteins, whereas chitinases contributed ca. 16 % + 2.6. For A. collinsii, glucanase proteins contributed ca. 60 % + 2 and chitinases contributed ca. 19 % + 1.6 (N = 3 gels representing biological replicates).

Activity of glucanases and chitinases in EFN
Native gel assays were employed to separate basic and acidic isoforms of the dominating protein classes, chitinases and glucanases, and to prove their activity. Active isoforms of acidic and basic β -1,3-glucanases and chitinases were highly represented in 210 the EFNs of the three myrmecophytes ( Fig. 4). By contrast, EFNs of nonmyrmecophytes exhibited mainly activity of several isoforms of acidic chitinases and one band representing an active basic chitinase (Fig. 4 a,b), but no bands representing active glucanases (Fig. 4 c,d).

DISCUSSION
Extrafloral nectar of myrmecophytic Acacia species appeared free of fungi and yeasts when collected under natural conditions, whereas the EFN of sympatric and closely related non-myrmecophytes was regularly infested (González-Teuber et al. 2009). The 220 purportedly biological activity of the nectars was investigated in bioassays on six phytopathogens, some of which have been previously reported to infect Acacia species (Roux and Wingfield, 1997;Kapoor et al., 2004). These pathogens could, thus, even use nectaries as entries to infect the plant tissue, as speculated earlier by Bubán et al. (2003) and Farkas et al. (2007). EFN of myrmecophytes inhibited the growth of at least four among these species, including all those reported as pathogens of Acacia species (Table   2), whereas no inhibitory effects were observed for non-myrmecophyte EFN ( Table 2 As the resolution of one-dimensional gels was too low to obtain a complete image of nectar proteomes, two-dimensional gel electrophoreses followed by MALDI-TOF/MS and LC-MS/MS were employed to characterise the proteomes A. hindsii, A. 240 collinsii and of the two non-myrmecophytes (the proteome of A. cornigera EFN has already been described, see González-Teuber et al., 2009). The 2D gel analysis revealed a much higher number of protein spots than the 1D gels (Fig. 3). PR-proteins such as chitinases, glucanases and thaumatin-like proteins clearly dominated the protein fraction both qualitatively (Table 5) and quantitatively: putative glucanase proteins contributed more than 50 % and putative chitinases more than 16 % to the total proteins in EFN of both myrmecophytes. These values are slightly higher than those identified earlier for The non-PR proteins found in myrmecophyte EFN were identified, for example, as glycoside hydrolase and glucan-1,3-β-glucosidase (Table 5). The functions of these 260 enzyme classes are commonly associated with carbohydrate metabolic processes (Zoran, 2008;Bojarová and Křen, 2009) and are likely involved in the control of the nectar's carbohydrate fraction, as has earlier been described for an invertase that is also present in the EFN of Acacia myrmecophytes (Heil et al., 2005;González-Teuber et al., 2009 Although SDS-PAGE analysis revealed very low abundances of proteins in the EFNs of both non-myrmecophytes (Figs. 2, 3), native gel assays indicated activities of several acidic and at least one basic chitinases in these EFNs (Fig. 4), whereas no clear hint towards active glucanases were obtained (Fig. 4). For the EFNs of the three myrmecophytes, the activity assays confirmed the analysis of 2D-gels with nanoLC-270 MS/MS (Table 5), since active isoforms of acidic and basic β -1,3-glucanases and chitinases were highly present in these EFNs (Fig. 4).
In summary, most proteins present in the EFN of myrmecophytes have as yet been reported in the context of the defense of plants against pathogens. Do these proteins fulfil the same function in EFN, which represents a secreted liquid rather than tissue, and does the protection of EFN from microbial infestation rely completely on PR-proteins? Size exclusion filtration and an experiment in which enzymes had been deactivated by heating confirmed that only the active protein fraction of EFN of the myrmecophytes inhibited phytopathogen growth (Tables 3, 4). Moreover, the pathogens tested in our bioassays (Tables 1, 2) included the oomycete Phytophthora parasitica and 280 different ascomycetes, which all differ in the composition of their cell walls. β-1,3glucans are the major component of hyphal walls of oomycetes, while ascomycetes have cell walls made of varying proportions of chitin and beta-glucans (Wessels, 1994).
Interestingly, the strongest inhibitory effects were found on the oomycete Phytophthora parasitica (Tables 2, 3). This result indicates that glucanases (the most abundant PRproteins in EFNs) were mainly responsible for the observed effects. Probably, the absence of active glucanases from the EFN of non-myrmecophytes explains the differences in the protection from microorganisms between EFNs of both functional plant groups (Table 2). Indeed, some in vitro studies showed that purified β -1,3glucanases exhibited activity against phytopathogens such as Phytophthora and 290 Alternaria (Tonón et al., 2002;Liu et al., 2009). Whereas the activity of plant protein extracts against different Phytophthora species lies in basic glucanase isoforms (Meins and Ahl, 1989;Kim and Hwang, 1997;Yi and Hwang, 1997), tobacco basic chitinase and glucanase isoforms exerted antifungal effects (Lawrence et al., 1996;Joosten et al., 1995;Sela-Buurlage et al., 1993). In vitro experiments have demonstrated that the antifungal effects of chitinases and β -1,3-glucanases are synergistically enhanced when both enzyme classes are present (Vogeli et al., 1988;Sela-Buurlage et al., 1993;Lawrence et al., 1996;Anfoka and Buchenauer, 1997).
In addition, PR-1 and PR-5 (thaumatin-like and osmotin-like proteins) have been associated with activity against oomycetes (Van Loon et al., 2006) and representatives 300 of both families were detected in myrmecophyte EFN (Table 5), but according to their molecular weights are not likely to be present in the EFNs of the non-myrmecophytes: the latter did not show protein bands below 20 kDa (Fig. 2). These PR-proteins have been associated with strong antimicrobial activities in vitro (Monteiro et al., 2003) and might, thus, further contribute to the inhibitory action of myrmecophyte EFN against filamentous microorganisms. All these observations suggest that the activity of PRproteins underlies the protection of EFN from microorganisms and that the high protection level that can be observed in the EFNs of Acacia myrmecophytes is caused by the combined activity of several different PR-proteins.

CONCLUSIONS
The present study allows a clear allotting of the antimicrobial protection of myrmecophyte EFN to its protein fraction and demonstrates that the nectar exhibits a significant effect against purportedly phytopathogenic microorganisms. Because nonmyrmecophytes secrete EFN only transiently at low rates and in response to herbivory,

Study species
The study was conducted using the three myrmecophytes, Acacia cornigera (L.) Willdenow, Acacia hindsii Benth. and Acacia collinsii Safford and the two non-

Antimicrobial activity of EFN
Bioassays were carried out to evaluate a putative inhibitory effect of EFN on the growth of six phytopathogens: the oomycete Phytophthora parasitica and the ascomycetes

Fusarium oxysporum, Verticillium dahliae, Alternaria alternata, Botrytis cinerea and 380
Plectosphaerella cucumerina. The genera Phytophthora, Fusarium and Verticillium have been previously described as leaf pathogens for other Acacia species (Roux and Wingfield, 1997;Kapoor et al., 2004;Chimwamurombe et al., 2007) (see Table 1). All species were maintained in potato dextrose agar (PDA, Sigma) plates. The antimicrobial assay was performed following the disk diffusion method: sterile filter paper discs (0.5 cm diameter; equidistantly separated) impregnated with 10 µL of EFN were placed on the surface of a new PDA plate. One agar plug (2.5 cm 2 ) containing growing hyphae that were obtained from the edges of an actively growing colony was placed in the center of the plate. Plates were then incubated at room temperature for 72 h. After incubation, plates were inspected for inhibition of microbial growth around the filter 390 paper disks. The inhibition zones indicate that the respective microorganism is susceptible to some activity present in the EFN. Inhibitory effects of EFN on the six microbial species were quantified as the percent of the line from the centre of the microbial colony towards the nectar droplet that was not occupied by the microorganism. The experiment was conducted with EFNs of all five species.
Myrmecophyte EFN (A. cornigera, A. hindsii and A. collinsii) was adjusted to a concentration of 10% (w/v) using a portable refractometer, which represents the common EFN concentration found in the field for those species. A 10% sugar solution (fructose : glucose, 1 : 1) was used as a control for the EFNs of these species because EFN of Acacia myrmecophytes is naturally free of sucrose (Heil et al., 2005). Non-400 myrmecophyte EFN (A. farnesiana and Prosopis juliflora) was adjusted to a concentration of 3% (w/v) and a 3% sugar solution (sucrose : fructose : glucose, 1 : 1 : 1) was used as a control. All assays were repeated with three biologically independent replicates for every microorganism.
To evaluate which fraction of EFN causes its putative antimicrobial effects, a membrane filtration of 5 kD (Vivaspin 500, Vivascience Sartorius Group, Stonehouse, UK) was used to separate the protein fraction (> 5 kD) from the metabolite fraction (< 5 kD) of EFN. After centrifugation (13.000 rpm for 5 min) both fractions were obtained for each plant species, and the disk diffusion method (see above for methodological description) was carried out using Phytophthora parasitica as the target organism. 410 As an independent attempt to causally relate the antimicrobial activity of myrmecophyte EFN to its protein fraction, EFN was collected from the three myrmecophytes in November 2009 and subjected to disk diffusion assays with the three fungi, Alternaria alternata, Fusarium oxysporum and Botrytis capsici. Half of the nectar was heated to 100°C for 15 min before placing nectar droplets onto the Agar Plates. The experiment was repeated with three biologically independent samples for every fungal species.

SDS-PAGE, Two-PAGE and mass spectrometry
Before one-or two-dimensional SDS-PAGE (polyacrylamide gel electrophoresis), 420 nectar samples were adjusted to a concentration of 3% of soluble solids (w/v) using a refractometer (10-20 µL for myrmecophyte species, and 150-200 µL for nonmyrmecophyte species), and then precipitated with 10% TCA (v/v) at 4° C (nectar : TCA = 1 : 2). The mixture was incubated for 1.5 h at 4° C and centrifuged at 13000 rpm for 15 minutes at 4°C. Then, the supernatant was removed and 0.5 mL of absolute ethanol was added. Samples were centrifuged at 7000 rpm for 10 min at 4° C. Proteins

In-gel assays of activities of glucanases and chitinases
Acidic and basic chitinases and glucanases were determined by native gel assays in order to detect and separate active isoforms in nectar. For the detection of acidic and neutral chitinases and glucanases, 10 µg of proteins per sample were separated by PAGE in 15% (w/v) polyacrylamide gels under native conditions, at pH 8.9 according to Davis (1964). Glycol chitine was embedded in gels at 0.01% (w/v) and used as substrate for chitinase activities. After electrophoresis, chitinase gels were incubated for 4 h at 37° C in sodium acetate buffer 50 mM, pH 5.0. For β -1,3-glucanase activities, a soluble fraction of purified β -glucans from Saccharomyces cerevisiae was used as a substrate (Grenier and Asselin, 1993). β -glucans were incorporated at a final 500 concentration of 0.6 mg mL -1 directly in the separation gels. After electrophoresis, glucanase gels were incubated for 3 h at 37° C in sodium acetate buffer 50 mM, pH 5.0 as well. Running conditions for electrophoresis of chitinases and glucanases were 100 V for 1.5 h. Chitinase activities on gels were revealed by fluorescent staining (10 min) using calcofluor white M2R (0.01% w/v) in 500 mM Tris-HCl (pH 8.9) and visualised after destaining under UV light. Glucanase activities on gels were revealed by staining the gels for 15 min with 0.025% (w/v) aniline blue fluorochrome in 150 mM K 2 HPO 4 , pH 8.6, and visualised under UV light (365 nm).
For the detection of basic chitinases and glucanases, 10 µg of proteins per sample were analysed by PAGE in 12% (w/v) polyacrylamide gels under native 510 conditions, at pH 4.3 as described by Reisfeld et al. (1962). For basic activities, substrates of glycol chitin and β -glucans were incorporated in an additional polyacrilamide gel (overlay gel, 7.5%) to which proteins were transferred. Transfer of proteins was done by blotting for 3-4 hrs at 37º C. After electrophoresis, separation gels (attached to a supporting glass plate) were covered with the overlay gel. Bubbles